Huong Nguyen Thi Minh
Khanh Vuong Tuan
Hoang Ho Duy
Tri Le Tien
Thang Nguyen Huu
Dien Truong Chi
Organization: FPT Software
FPT Corporation is a significant global provider of technology and IT services, with revenues of almost US$1.3 billion and 30,000 employees in 26 countries. As a pioneer in digital transformation, FPT provides services of the highest caliber in Smart Factory, Digital platforms, RPA, AI, IoT, Enterprise Mobility, Cloud, AR/VR, Business Applications, Application Services, and BPO, among others. The company has served over 700 customers globally, including 100 Fortune Global 500 corporations in Aerospace & Aviation, Automotive, Banking & Finance, Logistics & Transportation, Utilities, and other industries.
We work as an outsourced data team for a financial consulting company's securities trading department. This company provides students with a regularly updated market analysis tool in addition to training.
When we first launched the platform, we encountered several issues in acquiring a large volume of data from a variety of sources, and certain data requires real-time updating. The data must be managed so that other teams, such as the FA team (fundamental analysis - requires financial reports, news events, etc. from several sources) or the TA team (technical analysis - requires real-time updates based on chart movements), can review it and develop dashboards.
At the same time, these statistics are available for use in drafting reports, highlighting significant occurrences, and staying current with market news movements before handing off forecasting to the Predict team (results of both teams' needs analysis) for market and stock trends, and so on.
Since the analysis teams continue to operate independently and lack a centralized system for processing daily and hourly data updates, the data is not regularly updated, rendering the work of predicting trends or developing dashboards inaccurate and ineffective for students.
We'd heard of Dataiku and had two weeks to learn everything there is to know about the platform. Our initial impression of Dataiku was that the interface is quite user-friendly; that dataset management is straightforward; and that creating dashboards, modeling, and feature engineering are simple processes for performing data visualization. Furthermore, Dataiku's visual recipes have made ETL work simpler via recipes such as Pivot, Join, Prepare, Sync, and Split. As a result, we decided to use Dataiku to create a brand new platform for computerized data processing.
Our information is gathered from various sources, including articles, news items, listed company websites, and financial reports from stock exchanges. We can easily access big data sources like these thanks to Dataiku.
There was a lot of data that needed to be processed in real-time. As a result, we had to combine Streaming Spark, Streaming Python, and KSQL. Despite a slight lag, our analytics system has improved the accuracy of dashboards, reports, and so on.
The user-friendly platform of Dataiku allows us to systematize, easily regulate the data source, and directly control the data quality. All the work in this challenge can be managed by three to four people, including data collection and storage, workflow/schema consistency checks, ETL, modeling, and the automation step.
Additionally, data scientists no longer need to combine various tools such as Python and PowerBI to perform EDA, develop dashboards, and model on the same platform. Thanks to Dataiku, we no longer have to produce them independently. This is really practical and easy to use.
Business Area: Financial Services Specific
Use Case Stage: In Production
The teams refurbished significant analytical systems at the stock market teaching facility that generated daily, hourly, and continually updated stock market analysis reports throughout their first eight months with Dataiku.
This project is a big accomplishment, automating time-consuming spreadsheet activities, boosting the volume and frequency of analytics, and delivering self-service analytics capabilities in a regulated, standardized manner.
Furthermore, when the analysis reports are constantly updated on a regular basis, the center's pupils can make trading decisions in the stock market in a safe, less hazardous manner. Specifically, with FA analysis, reports are visualized by industry, making it simple to find the best company that meets specific criteria. With TA analysis, meanwhile, students can easily understand market movements, such as the most traded stocks, which groups of stocks are of the most interest, or which trading indicators are updated instantly, giving them a competitive advantage in the market.
With the values brought by the analysis system mixed with the center's curriculum, the number of students at the center is steadily expanding rapidly, as is the center's reputation. In the most recent quarter, the number of pupils increased by 23%. In the future, we also want to develop a trading simulator so that trainees can gain extensive experience before entering the market.
In the past, it took our analysis team hours to update the visual reports for students. Thanks to Dataiku, the reports are now updated automatically.
Our financial analysts take only five to ten minutes to inform students on the latest news and produce market forecasts that are more accurate because they are automatically and regularly updated. As a result of the new system's enhancements, we lowered the number of analysts from 20 to five while increasing productivity by 20%.
In teaching, when there are continuously updated reports, more practical lessons are connected, and the number of hours of theoretical lectures is reduced by 30%, which increases the practice time. As a result, teaching efficiency is improved, and students feel more satisfied.
In addition, Dataiku provides data architects with the assurance that analysts are working with high-quality data, owing to the transparent visualization of data flows. Thanks to Dataiku's easy visual interface and process automation, our team was able to complete data preparation tasks efficiently. Dataiku also enables data scientists to interact directly with analysts' data to apply ML approaches to other business aspects more effectively.
Value Range: Thousands of $